About this site
Here we'll review recent developments in drug discovery and medicine and the IP issues and financial implications they have, along with general thoughts about research. Also likely to make an appearance: occasional digressions into useful topics like which lab reagents smell the worst.
About this author
Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases.
To contact Derek email him directly.
A Thundering Herd
A good bit of blogging time tonight was taken up by another worthy cause - a spectacular aurora, by far the most impressive I've ever seen. It seems to have quieted down for now, but you never know about these things. From the looks of our sky, this must have been visible well down into the southern states. I live in the Northeast, and we had a display nearly from horizon to horizon.
I did want to mention something unusual in the drug industry, though. Novartis is gearing up to staff their new facilities in Cambridge (which I spoke about back in the spring here.) They've done something that I've never seen a drug company do before, and I'll be interested to see how it works out. A recent issue of Chemical and Engineering News was mailed out with a special insert in its plastic bag: a card from Novartis.
It started off: "You are cordially recruited. . ." and went on to invite everyone to a large meeting, where representatives of the company would speak about employment at the new site. Now, I don't know how many people they sent this to - the entire membership of the American Chemical Society? Surely not. Maybe just those who were listed as members of the Organic or Medicinal sections? Maybe just subscribers in the East? I'd be interested in hearing from folks in other parts of the country.
This fiesta takes place on a Thursday, which is an interesting choice. If it had been on a Saturday, they'd have had a cattle call for sure. But a weekday will fetch only those people interested enough to take a day off from their current job. Of course, the other companies in the region will no doubt be watching very closely to see who takes that vacation day, won't they? I've even heard speculation that some might send along a ringer or two just to, ahem, take a head count of who shows up.
Well, I won't be there, myself. I'm not looking to move companies (although I've done it before, and it's often the best thing you can do for your career.) Now, if Novartis wants to triple my salary, then I'd be glad to take their call. Or if they want to pay me just to blog and come up with crazy ideas!
More Memantine, More!
Fellow Corantean Zach Lynch had an interesting response to my post about memantine, the Alzheimer's drug. He makes some interesting points, but in the end, I think that he and I are going to have to agree to disagree on this one.
Zack mentions an improvement in cognitive performance in some patients, but it's important to remember that these are measured versus control groups, who are presumably still deteriorating. The magnitude of the effects, at any rate, aren't enough to convince me that there's anything going on outside of "merely stopping or slowing damage." (By the way, that's a pretty hefty "merely." Anything that could truly stop the damage involved in Alzheimer's would be a breakthrough indeed.)
He goes on to ask "why not consider the evidence that memantine is working through a more subtle mechanism than preventing exitotoxicity?" I guess that my answer is largely based on Occam's razor. The NMDA actions of memantine seems central, and its effects must be peculiar to a diseased neuron. After all, giving the drug to normal volunteers actually impairs their memory, as you'd expect an NMDA antagonist to do. Reversing some sort of low-level, tonic excitotoxicity is the first explanation that comes to mind.
But don't get me wrong - because it's the easiest explanation doesn't mean that it's a simple one. Relieving tonic excitotoxicity is a pretty subtle mode of action all by itself, because there have to be some subtle differences in the compound for it to even work. Consider that memantine is clinically tolerated, while the stronger competitive antagonists (like MK-801, or even PCP) are (to put it mildly) contraindicated with enough preexisting dementia of their own.
And keep in mind that the receptor itself is quite complex. NMDA receptors have seven distinct subunits, and a wide variety of small-molecule binding sites and binding modes. There are layers of complexity on top of that, since they occur in different types of neurons (where they presumably have somewhat different functions) and even in different places on the same neurons. (There's evidence that synaptic and nonsynaptic NMDA receptors function rather differently, for example.) Memantine's actions have been intensively studied, but there's still a lot of detail we don't know.
Actually when Zack advances the theory that in Alzheimer's, "NMDA is overactive, letting in any memory at all and promoting the cellular changes that allow memories to be stored at random," I think he's actually talking about the same tonic excitatory action that I am. I'm not sure about the random-memory part of the idea, although that's an interesting one. For one thing, I don't know if the sort of low-level excitation is the kind that would give the memory-making long-term potentiation process a leg up or not - like any ion channel, the NMDA receptor has quite a few signaling modes.
And even if it is additive, I'm not aware of support for Zack's one-memory-after-another model. I would think that the second-messenger system downstream of the receptor would get depleted by such treatment - or if not, then there should be some things that are upregulated in order to keep up with the constant signaling. I haven't seen any evidence for that, though. I lean more toward the idea that the chronic stimulation actually interferes with normal NMDA action in a way that makes long-term potentiation (and memory formation) too difficult, rather than too easy. But it is an interesting take, and I'd be glad to hear from anyone with some corroboration.
Past NMDA, there are other possibilities. Perhaps memantine's activity at another ion channel, 5-HT3, is part of the story. There are recent reports that the compound affects all sorts of things in the Alzheimer's etiology, from amyloid formation and release through tau protein phosphorylation. These are suggestive results, but I wonder how many other compounds (especially ones without efficacy) have been looked at in the same way - how common are these effects? Also, I can't help but think that a compound that affects so many key Alzheimer's pathways would have to be more efficacious than memantine!
It's not until his last paragraph, though, that Zack actually loses me. "Complicated explanations may not suit a pharmaceutical company, but understanding the true therapeutic nature of current pharmaceuticals will be an important part of future neuroceutical development." No argument with that last part, but as for the first - well, complicated explanations are actually fine with us in the drug industry. For the most part, they're all we have, because nothing ever seems to be simple. And as I've pointed out above, I think that Zack's hypothesis suffers in some ways from not being complicated enough.
Nope, understanding the true nature of current pharmaceuticals suits us just fine. We don't find 'em under cabbage leaves, y'know. The problem is, I don't think we know the true nature of anything in biochemistry or molecular biology yet - we just have to keep at it and get as close as we can with the tools we have. Especially in the area of drugs for the brain, I think it's going to take a while, and that's probably where Zack and I differ the most. His timetable for the development of new classes of CNS drugs is a lot more accelerated than mine - but, then, none of the ones I worked on ever panned out.
But that's not to say that I don't think it can be done. I just think it's going to take longer, with a lot more fits, starts, and blind alleys than everyone would like. So in a wider sense, Zack and I are on the same team, especially when you compare us with people who don't think that such things can ever be done at all. Or with people who haven't even starting thinking about it yet!
Those Were Sort of The Days
Not long ago, I was digging around in a drawer upstairs when I came across a shirt that I thought my kids would be interested in. It's a dark blue cotton T-shirt, nothing special, except it has a slash of oval holes running across the front, flanked by smaller dark spots where the fabric was partly eaten away. "This," I told them, "is a shirt that I used to wear when I did chemistry in school."
There are several things about it that are different from what I wear today. First off, I have to admit that it's a bit small. My poverty diet in graduate school shrank me to about as bony as I've ever been. Second, well, there are those holes. Those came from a blast of concentrated sulfuric acid, which I lackwittedly poured down the sink one day from an unlabled flask. It hit the cold water in the trap, then did what con-sulfuric does when it's mixed with too small a volume of water. It popped, hissed, and splattered right back up, splashing me in the process. Realizing my mistake, as even the most thick-witted chemist would by then, I stripped off the shirt and began splashing myself with some saturated bicarbonate solution. I kept the shirt as a souvenir.
These days, I don't wear T-shirts to work. At the time, I had a whole wardrobe of them, and wore them until the dead of winter. But I was already switching over to what I've worn since (various button-down shirts, with the sleeves partially rolled up, worn with various sorts of khaki-like trousers.) That'll get you by in basically any industrial lab in the world, in my experience. The battered T-shirt look tends to remind people of graduate school, often not a source of pleasant memories for scientists. (And all this talk of clothes reminds me that I've followed the custom of my fellow Coranteans and added a photo to my biographical note at left. Now anyone who sees me at a meeting can tell me what they think of the site in person. Hm.)
On a different scale than sartorial, I haven't made that particular acid-pouring mistake since, fortunately. For one thing, I don't go heaving things down the sink with abandon. We did that a lot twenty years ago, but I think it's frowned on even in academia these days, with reason. And I'm not foolish enough (any more) to keep things like concentrated acids around in unlabled flasks, either, a lesson that this incident certainly helped to drive home. I can look back on that accident and shake my head a bit - at my inexperience, at my style of working. But I shouldn't be too complacent, because I can't begin to say what work habits of mine I'll be shaking my head about twenty years from now. . .!
Late last week, the Lancet really unloaded on AstraZeneca, on its CEO, and on its new statin drug, Crestor. (Said CEO has an understandably combative reply published in the same issue.) Now, the British medical press is a lot more lively than its American counterpart (in the case of the rival British Medical Journal, it's so lively as to be an arguable liability.) But this piece really leans on the horn, by their standards or ours. It's already made the British newspapers, and I doubt if the story's over yet.
Here are a few samples: "With no clinical end-point trial yet completed, the company has chosen to market rosuvastatin by applying adventurous statistics to an overinterpreted syllogism. Unfriendly. How about this part: "Physicians must tell their patients the truth about rosuvastatin--that, compared with its competitors, rosuvastatin has an inferior evidence base supporting its safe use. " And they really air their feelings about AZN's clinical data: "It is difficult to understand how such blatant marketing dressed up as research can appear under the name of a respected peer-reviewed medical journal."
Well. The Lancet's editorial staff clearly needs to try to get in touch with their feelings; it's unhealthy to keep things bottled up like this. But actually, they probably have a point. They observe, correctly, that AZN has a huge amount riding on the success of Crestor, and it's easy to imagine that the company has been pushing things as hard as they possibly can. Most of their clinical data seems to be derived from fairly short trials, and the results have been hyped for all they're worth, as clinical results usually are. \
And as the editorial points out, the big question is whether Crestor has a Lipitor-like side-effect profile, or a Baycol-like side-effect profile. Bayer withdrew the latter drug because of problems with rhabdomyolysis, and AZN has already backed off their highest dose of Crestor due to safety concerns. Other statins show the same problem, at varying rates. The question is always how wide the margin is between efficacy and side effects. Lipitor came in as a very potent statin, and has prospered without too many problems. Baycol was potent as well, and ran into trouble (Bayer's still digging out from under the mountain of lawsuits.) Crestor is being marketed for its potency - same effect at lower dose! So what happens when millions of people start taking the drug for an extended period?
The Lancet's position, boiled down, seems to be that there is no unmet medical need here, that the existing statins do about all that a statin can do. (And there's still a lot of room to argue about just how much that is - does extended statin treatment really lower mortality rates, or not? It's a question you'd think would be more worked out than it really is.) That stipulated, they're arguing that there's no point in exposing patients to a drug that will do them no extra good, and possibly some harm.
And they're certainly within their rights to argue that, and physicians are within theirs to refuse to write for Crestor if they're not convinced it has enough benefit. It's not like they'd be depriving their patients of the only statin available. This, to me, is the free market at work. I'd rather have the physicians doing this sort of thing than the insurance companies, any day.
This Coyness, Rodent, Were No Crime
Just a short piece tonight, with extra credit points to all the liberal-arts science types who get the source of that title. I'm catching up here on Sunday night, getting ready for another week at the Wonder Drug Factory. Longtime readers of this blog know that I don't spend too much time talking about day-to-day stuff from my work, for obvious reasons. It's a pity, because a blogged drug discovery project would make for interesting reading. A little too interesting, though, from an intellectual property standpoint!
What I can say is that right now we're involved in an odd sort of problem. We have too many good compounds. I know, I know, some of my industrial readers want to throw food at me right about now, but you guys know what I'm talking about, too. We have a whole list - a long list - of compounds that all look just fine in all the assays we have set up.
And that's the problem. There isn't a good way to differentiate all these nice compounds. We could, had we but
mice world enough and time, just scale 'em all up and put them in the animal models. That would sort the lot out, I'm sure. But just the thought gives me the shakes: all that time, all that effort, all those compounds, all that work in the in vivo group. Not possible, not even remotely.
So the hunt is on for an assay that will spread things out a little. It would be nice if it were a relevant assay, but projects have been known to waive that requirement in moments of desperation. Let's hope it doesn't come to that.
A few short follow-ups to recent posts. . .first off, yesterday's. A former colleague of mine writes to point out that chromum oxidations often generate green reactions. He's absolutely right. It's something that I should have remembered, since I once had to scrub the residue of one of them out of my hair. (That's a story from my Lagniappe blog archives, which currently seem to be hosed. I'm working on a long-term retrieval project for those.) All this has reminded Anthony Cox over at Black Triangle of the "Blackadder" series, and it wouldn't be the first time my chemistry has fit right into a British comedy.
Chromium has a number of spectacular colors. The +6 oxidation state compounds tend toward warning-label oranges and reds, and the +3 oxidation-state ones can be green, or (in the case of the chloride) an extraordinary purple. That one looks exactly like something that should be on the body of a 1960s drag racer; it's a rich, glittery heap of metallic purple flakes. Try to make an organic compound look like that! I've wished for years that it had some plausible use in my synthetic chemistry, so I could order a big bottle of it just to have around.
Also, by mentioning the holding-the-flask pose common to brochure chemists, I didn't mean to slight other well-known poses. For biologists, it's Thoughtful Pipetting, over and over, and there's always the classic Thoughtfully Peering Into A Microscope. If anyone tried to be as continuously thoughtful as people appear in this propaganda, they'd sprain their head.
No fans of Forest Labs wrote to take issue with my opinions on memantine. But I should note that today's Wall Street Journal says that the drug could end up being a big seller (and they may be right, given the size and desperation of the market.) The article suggested that FRX stock might well be a buy, in the teeth of the wind that's pushing most drug stocks back to port. I don't know if I believe in that idea enough to act on it, but if I do, I'll let everyone know.
Which reminds me, I'm still short Imclone at $40 per share. An analyst has been pounding the table about their prospects recently, but (oddly enough) he's bearish on them, too. His take is that IMCL should be down in the 20s, and I have to say that we have exactly the same take on the stock. That should comfort me a bit if it zips up to $6 or something. Wouldn't be the first time that happened with something I'm short.
And while I'm mentioning stocks, this is of course earnings-report season. Just after I'd been talking about Merck having a more realistic earnings-growth target than Pfizer, they came out with even gloomier figures. (But over on Planet Pfizer, everything remains just fine, apparently.) Merck is cutting several thousand jobs, although I've yet to hear how many will be in research. Early retirement plans are one thing, but when a drug company starts truly eliminating research jobs, you know that there are some serious problems. Anyone at Merck have any details?
Join The Vanguard, Jolly Lab-Comrades!
I had the occasion this week to run a large reaction that (at least at first) is a bright, pure green. Green is an unusual color in chemistry, and especially this lime-jello shade. It's nice enough that I took the flask and wandered up and down my hallway showing it off. (I also knew that the addition of the next reagent caused the whole thing to turn black, which isn't quite so impressive!)
When you get a color like that in an organic chemistry lab, it's almost always because you're using some transition metal. "Check this out!" I called to a colleague today (the reaction had developed a white precipitate, and was now the color of mint-flavored ice cream.) "Nice," he said. "Copper?" "Nickel," I replied. Those two account for pretty much all the greens that we're ever going to get.
Another accurate name for the shade would be "annual report green." That's to go along with the similar reds, blues, and so on that feature prominently in those entertaining documents. (And they're usually shown in side-by-side flasks, as if one of those shades weren't unlikely enough on its own.) And often as not, one of the flasks is being held up by a hard-working, far-seeing researcher in an attitude of intense concentration. Breakthroughs are always a loud primary color, of course.
I call the style of these photographs "Corporate Realism," to go along with the well-known socialist type.
Might I add that it's about as successful, and about as appealing? Fellow chemists, next time management sends the photographers around, next time they're unloading the baby spotlights and the colored gels to bounce off the reflectors (for the ever-popular spash-of-red-and-blue-light in the background,) strike a blow against the ordinary! No more flask-holding shots! No one cares if they never see another one again! Hold up something different next time - or if you have to hold up a flask, hold it up like you usually do. You know, furrowed brow, pursed lips, the whole what-the-hell-happened-this-time expression. Now that's realism!
A New Drug for Alzheimer's?
Forest Labs has been making the most of their recent FDA approval for Namenda (memantine,) and I suppose they're entitled to. This is the first Alzheimer's drug to make it onto the US market in years, and it's aimed at a large and cruelly underserved patient population. But as someone who did research in the field for several years, I find it hard to get very worked up.
Chemically speaking, though, I like the compound a lot. It's a simple derivative of adamantane, a molecule that I can only describe as cute. Adamantane is a basic repeating structural unit of diamonds. It's like four six-membered rings of carbon fused together in the tightest 3-D way possible. I have a spot in my heart for pharmaceutical structures like these; I think they're underused. So while my drug-industry side isn't too impressed, the pure chemist in me wishes the compound (and its relations) well.
Memantine has the virtue of novelty. It has a different mechanism than the other existing AD drugs, which are cholinesterase inhibitors. The best known of those is Pfizer's (actually, Eisai's) Aricept (donepezil.) The cholinergic neurons are known to take a lot of damage in the course of the disease, and these drugs attempt to get the most use out of them while they're still functioning. Cholinesterase is the enzyme that clears a particular neurotransmitter (acetylcholine) from the synapse after its job is done, and inhibiting the enzyme strengthens and prolongs the signaling. This mode of action is basically tries to jolt a system that's already breaking down in an Alzheimer's patient. It's a pretty crude approach, but it's all we have.And that explains why the cholinesterase inhibitors don't work that well, or for that long. They'll slow the progression of the disease, particularly in its early stages, but they're fighting a losing battle. There is no known method to actually stop the course of Alzheimer's, and certainly no way (yet) to reverse the damage it causes. (That damage, to tell the truth, is probably irreversible after a certain point.)
Memantine's a different beast. It has some affinity for a wide range of receptors in the brain, but at the doses that are seen therapeutically, the relevant interactions seems to be with the NMDA receptor. This is a ligand-gated ion channel with a huge research following in the CNS field. It's a complex receptor, with a lot of distinct binding sites, each with its own function and own intricacies. I could not begin to summarize the amount of work that's been done on it over the years, although it would probably do me good to try. The connection with Alzheimer's is that NMDA-containing neurons are known to be key players in memory formation and retrieval.
The weird thing is, memantine is an antagonist; it blocks NMDA signaling. So at high doses, it actually interferes with memory. (Other NMDA antagonists do the same thing, which is one of the pharmacological lines of evidence that led to studies of the receptor's role in such processes in the first place.) Actually, NMDA antagonists as a group are a pretty rough bunch of compounds. Consider, of all things, PCP and ketamine. Not the first place you'd look for a well-tolerated drug! Antagonists aren't all alike - there are clearly some important differences between the actions of memantine and of notorious compounds like those two.
At clinical doses, the compound does play against type and seem to improve memory. The best guess for how this works is through a mechanism for neuronal injury in Alzheimer's. Too frequent (and too prolonged) firing of excitatory pathways like NMDA have long been associated with cellular damage in the brain, and this seems to be going on in AD as well. Memantine appears to be the right sort of antagonist to slow this process down, taking the tone of the NMDA system back toward the normal range.
Unfortunately, this interesting mode of action doesn't seem to translate to anything dramatically different, clinically. Memantine certainly seems safe enough - it's been used in Germany for years now. But at best, it just slows the progression of Alzheimer's, for a while. That seems to have been one of the main topics of debate at the FDA, whether the drug had enough benefit to warrant approval. In the end, since patients with Alzheimer's have so few other options, they went with it. And it will help some people out - to a degree, for a while. That, presumably is worth something. I just hope that no one has their hopes up too high.
Pfizer and Reality
Looking through the news headlines for drug companies can be instructive, but not always in the ways that the companies were thinking. Whenever there's big breaking news on a poor outcome in a clinical trial, for example, you can bet that there will be headlines from the wires and the financial sites saying "Bad News For XYZ Corp." or "Drug Disappointment Sends XYZ Into Tailspin." And right next to those will be XYZ's own spin, released on that river of fantasy known as PR Newswire. The phrases will be as well-worn as the ones used by college football coaches. "We're committed to this drug. . .still analyzing the data. . .positive effects seen. . .subgroup analysis. . ." You know.
For example, here's a wire-service report on Pfizer. Their CEO gave an interview to the Sunday Frankfurter Allgemeine Zeitung, which is roughly the New York Times of Germany. The focus of the article is on Bayer, but the main news item was: no big mergers or acquisitions in the near future. So Pfizer's standing pat, and doesn't need to go outside for any new revenue streams.
Contrast that with this interesting table from Forbes. These are companies with major patent expirations in the next few years, and who leads the pack? Pfizer, indeed. Over half their blockbuster-derived revenue this year is going to disappear by 2008. Merck has similar problems, to be fair about it.
But check that table again: Merck is forecasting about 8% earnings-per-share growth during that period, the lowest of the companies on the list. And Pfizer? Pfizer says that they're going to shoot out the lights. They're going to show 15% annualized EPS growth, while losing huge moneymaking drugs from their portfolio. And they're not going to do any more big deals to make it happen, because their CEO just told us so
So all this ripsnorting growth, mighty revenues that'll make up for those patent losses and more, is going to come from their own current portfolio. Ve-ry well. It seems that everything is going to work, nothing's going to fail in the clinic or get withdrawn from the market (none of that Trovan business! They've learned their lesson!), and everything's going to sell like it's eternal life in a jar. . .hmm. . .
More "How Not to Do It"
When I was in graduate school, I thought about how I could assemble various techniques I saw into a useful lab handbook, with the title above. On my previous blog, I've already written about such things as How Not to Make Pyridinium Chlorochromate (and that title is as truthful as it can be; I witnessed what must have been about the worst attempt in the history of synthetic chemistry.) I could have produced a wide-ranging work, covering everything from How Not to Rota-Vap to How Not to Set Up the Hydrogenator.
I'd have chapters on such classics as How Not to Cut Sodium and How Not to Quench Your THF Still, but those topics have had ample coverage in the literature. A good eighty per cent of lab stories about fires start off with the phrase "We had this solvent still. . ." Last time around, I went into How Not to Handle Lithium Aluminum Hydride (that post has links to previous installments,) but that topic has also been covered by many regretful researchers. No, the real value of this work would be looks at such neglected subjects as How Not to Distill HMPA, How Not to Handle Hydrogen Bromide Cylinders, and How Not to Assign Five Water Peaks in Your NMRs.
In a more perfect world, I would have video footage of the original lab work to illustrate these points. Of course, in a more perfect world, these mistakes wouldn't have been made in the first place, but that way lies madness: to start with, in said world, I wouldn't have been working on half the things I did in graduate school at all. As it is, though, I have only my memory to rely on, which isn't always reliable (especially after, as of this fall, twenty years. But in my defense, I'd add that my recollection of some of these is, let's say, abnormally vivid.
Take that HMPA distillation. My colleagues in organic chemistry will know this solvent well. It's about as polar as a solvent can get, and does very interesting things to a lot of anion reactions. But its polar nature means that it mixes freely with water, and it soaks it up quite well on standing. You really have to dry the stuff, and the usual way is vacuum distillation from some dessicant (calcium hydride, in my experience, which come to think of I haven't used in at least ten years.) HMPA is also rather toxic, and we avoid it in pharmaceutical synthesis for that reason. No one would ever scale up a synthesis that relied on it, so what's the point of seriously using it? Academic chemistry, though, finds more uses for it, nasty or not.
One fine afternoon in graduate school, I was peacefully advancing the cause of science when one of the guys from down the hall came into my lab. "What's HMPA smell like?" he asked. "Holy (excreta)!" I answered, "You think I know? Probably like it tastes." He told me that one of our recent postdocs was distilling it, and he was afraid that the smell in the lab was, well, HMPA, odd as that might be.
I went down the hall to investigate, and came upon the single stupidest distillation rig I've ever encountered. There it was, a two-liter round-bottom flask with a heating mantle on it, boiling and bumping away on the high vacuum line. (OK, fair enough, if you're going to distill the stuff, might as well get it over with.) On top of this lunking load of toxic solvent was the smallest still head in the group, a tiny little 14/20 short-path job that looked about the size of the cherry on top of a triple banana split. This thing wasn't even slowing the hot HMPA vapors down much. My friend had a right to be suspicious, and yes, that smell probably was the springtime-fresh aroma of HMPA itself.
Unfortunately, I can't say much about the bouquet of this caricinogenic substance. You'd have to track down our Spanish post-doc and ask him; he was basically showering in the stuff. I stayed out in the hall while I ranted at him, and as he informed me that this was how they did it in Barcelona. "Well, go to Barcelona and do it!" I shouted, looking around to see if there were drops of solvent starting to run down the walls yet. I'll leave the subject by asking everyone to remember a key motto of science and engineering: You can't make things foolproof, because fools are too tricky.
From the Sports Desk
I had a call today from a local TV station, wanting to interview me. This isn't a regular feature of my life, so I was naturally curious. Turns out that they were looking for people who had the same names as members of the Red Sox or Yankees! I did a few minutes with them in a parking lot across the way from my workplace - I told them that the company tended to get a bit jumpy when they saw the TV vans pulling up, and they had no problem understanding.
One of the things I mentioned was that I wouldn't mind have a few per cent of my namesake's salary. Medicinal chemistry is a pretty good living, but I don't know many people who've become rich from it. Now, if US companies followed the German system, we might be on to something. Over there, I'm told, companies pay out a percentage of profits from a given patent to the inventors. So if you're on a real winner, it means a substantial raise for the life of the patent. That must lead to some really vicious fights over who was a real inventor and who wasn't.
But that led into another point. I told the TV interviewer that I would have a much better chance of training Derek Lowe, the pitcher, to be a medicinal chemist than he would have of teaching me how to throw a 95 mph slider. I don't think that my doppelganger would be all that great at chemistry (but who knows?) But it's for sure that I would stink mightily as a major league pitcher. It's such a rare set of physical skills - if I'd started practicing when I was six years old, perhaps by relentless effort I could have made it as far as single-A ball. I'd say that's about as high as an average physical specimen can make it on effort alone.
All this reminds me of a passage from G. H. Hardy's A Mathematician's Apology:
We have of course to take account of the differences in value between different activities. I would rather be a novelist or a painter than a statesman of similar rank; and there are many roads to fame which most of us would reject as actively pernicious. Yet it is seldom that such differences of value will turn the scale in aman's choice of a career, which will almost always be dictated by the limitations of his natural abilities. Poetry is more valuable than cricket, but Bradman [a great player of the time-DBL] would be a fool if he sacrificed his cricket in order to write second-rate minor poetry (and I suppose that it is unlikely that he could do better.) If the cricket were a little less supreme, and the poetry better, then the choice might be more difficult: I do not know whether I would rather have been Victor Trumper or Rupert Brooke. It is fortunate that such dilemmas occur so seldom.
I'll leave on that note, and return to seeing whether the wealthy Derek Lowe and his teammates will prevail. . .!
Off the Air
A power outage here has caused some household problems and eaten my blogging time for this evening. More from Pipeline HQ for Friday morning! All of you readers out there in the industry, go set up an extra experiment or something (I can tell from my hit counter that you're reading this site during working hours!) I promise to get something done in the lab, too, just to make it even.
And no, don't just go to an extra meeting and figure that you've done your part. It takes about five hundred meetings to equal the good that comes of one well-run experiment.
Plugs, Dept. of Shameless
I'm including some Amazon links to books that I think would be of particular interest to readers of this site. All of 'em are personally approved; the last thing I want is to visit my own page and find Amazon's robot shilling away for the "Total Idiot's Guide to Homeopathy" or the like. I figure that thousands of people will line up here to do all their shopping, and I can sit back and watch those nickels roll in. Book selections will change as the spirit moves me. Hope you find them interesting!
Picking and Choosing
Which projects are worth working on? I'm going to answer that in a very narrow sense: which ones would I want to work on myself (or have my lab assigned to?) I'm rating things on their chance of success, their potential for interesting science, and (negatively) their potential for wasting everyone's time. Today's post could serve as a brief guide for medicinal researchers, and especially for those new to the field.
I went into Rule Number One yesterday: avoid "sure things" at all costs. There are no sure things in drug discovery. The danger is that some managers above you might not know that, or might have persuaded themselves that this time is different. It isn't. If you get a sure-bet project to work, you will receive scant credit for it, and if it fails, it'll be your fault. You've been warned.
Rule Number Two is: avoid projects that have multiple "and then we get lucky" steps. All drug discovery projects depend on having some luck. But there's a limit. You can start off on one that requires you to find a much more potent compound, and has no clear route to finding one, sure. Maybe you will, and everyone will look good in the process. Or you can start one that requires you to overcome some black-box tox problem. That's a toughie, but it can be done. But don't start one that makes you do both of these at the same time. You don't estimate the probability of success by adding these things up: you have to multiply them together. Don't put yourself in the position of having a long-shot chance at taking a long-shot chance.
(For me, that rule means that I avoid central-nervous-system targets whenever possible. I did my time in CNS, eight years worth, and that's enough to know just how hard a field it is. Drug discovery is tough enough without having to get through the blood-brain barrier, and once you're in, it's not like we understand what's happening up there. Stay out in the periphery, is my advice.)
Rule Number Three: become the expert on your project. I've spoken about this in other posts. It's fatal to assume that everyone else is completely on top of the situation. Don't assume that your research organization wouldn't launch a project with some glaring flaw in it, causing the whole thing to eventually burst like a malign pinata. They just might. It's up to you to do your homework: if you're a biologist, keep an eye on your chemists. If you're a chemist, learn the biology. Don't go around taking everyone's word for everything.
That goes for both the positive and negative stuff. Don't take it on faith that a loser project is really a loser, not until you've checked it out yourself. Great credit goes to the people who make these things work; they're worth some extra effort.
Those are some of the rules, all of which I've learned through experience. My readers in the industry will be able to cite many examples that back these up, or they could if they were able to talk about them. Suggestions are welcome for extending the list - maybe we can come up with a project guide that would actually help people out.
Too Easy, Or Too Hard?
I've been on drug projects where we couldn't buy a good compound. The whole thing would start with a lead that was just barely good enough to get people to work on it. Then everyone piles in hoping to get something better. But sometimes that doesn't happen. You can end up trapped on this little island of structure-activity space, unable to make any major changes without dropping off the edge. (We try to do some preliminary chemistry to make sure that we're not stuck like this, but you can get fooled.)
And I've been on projects where it's like someone just shook out a big sack of good compounds. Nanomolar this, nanomolar that all over the place. One big long series of compounds after another, all of which work just fine. So which of these projects would I rather work on?
Weirdly enough, it's the first case. The problem with the other one is that you can end up with so many good compounds that it's hard to distinguish them. The expectation is that you're going to pick the best one to move into development, and what if you can't figure out which is the best? If you have a long list of potent compounds that all act about the same, then you need some sort of tiebreaker. Take a whole list of them into toxicity screening? That's not going to be popular - suddenly this project that everyone thought was going just fine turns out to be a big drain on resources. Alternative rankings tend to elevate trivial properties beyond their real importance, which makes people wonder why you're ranking compounds based on things that no one else cares about.
And the other problem with the cornucopia-type projects is that, invariably, the goodies don't shower down on every lab equally. It's no fun to be working on the section of the molecule that doesn't perform, when everyone else is cranking out the active stuff. (But the people making the active stuff complain, too, that their great compounds have been devalued by all the others.)
No, I'd rather be on the tough ones. That way, you have a free hand to try all sorts of odd ideas (you're desperate, right?) And if you actually get something to work, then everyone looks heroic. Meanwhile, if your lab isn't getting anything good to happen, hey, you've got plenty of company. Can't be your fault!
This is part of my longstanding preference never to be on sure-thing projects. That's because there are no sure things in the pharmaceutical industry, try as we might to ignore that fact. A project that everyone thinks is going to deliver has nowhere to go but down. If you deliver the goods, with ribbons and jingling bells on 'em, no one will care. You were expected to do that. Give me a marginal project - but not so fast. I don't want just any marginal one. I'll explain some of the criteria tomorrow. . .
Make Mine a Double
So, on Friday I wrote about "me-too" drugs, and whether there are too many of them (the majority view) or too few. I think the authors of the Nature Reviews paper have a point, but it's one that's vulnerable to a reducio ad absurdum argument.
That is, I'm willing to grant that the financial advantages of being first-to-market have been exaggerated. And I absolutely agree that it's often the second (or third or fourth) drug in a category that ends up making the most money. But what if everyone just decided to do me-too drugs? That would be insane, clearly, because all we'd see from then on would be incremental improvements in existing drug types. That would be enough of a problem by itself. But the whole me-too strategy would then burn itself out as it became harder and harder to improve on the state of the art for each kind of drug.
So we're not going to see that, for a lot of good reasons. The optimum research portfolio lies somewhere between all-originals and all-ripoffs, and we can get down to arguing about how much of each one we need. I think it varies from company to company. An outfit that's relatively flush with cash should probably plow a greater proportion of it into new ideas, since they're riskier and more expensive. Other companies with tougher prospects should bias their efforts toward improving some drugs that are already out there. They have a better chance for a higher return on investment.
But if your ROI is better for rip-offs, why innovate? We're getting close to a tragedy-of-the-commons problem now - someone has to come up with new agents that no one's tried before, after all. Several factors keep the innovative projects coming: company size is one factor. A small company will start based on someone's new idea, and the whole focus of the place will be on getting it to work. If it actually does take off, everyone does very well - look at Richard Gayle's old company, Immunex, for a good example.
Secondly, we've ignored the fact that companies can team up on newer or riskier kinds of drugs. Often this is a small-company big-company deal, with the bigger outfit taking a flier on something that might just work, meanwhile offloading some of the risk. And this applies to me-too drugs as well - they don't all come from other companies, as a glance around the marketplace will show. Interal ripoff projects are rife - some are big successes (Nexium for Prilosec) and others don't measure up (Clarinex for Claritin.) But it's an attractive strategy.
Maybe that's the answer: innovate, but prepare your own improved version as soon as you can. Traditionally, a company will time it so that its second-generation drug hits the market not long before the first one is losing patent protection, to try to maximize the first one's revenue. But that gives your competition time to work on their own revisions. Perhaps the best way to do it is to jump the gun, to get cracking as fast as possible on your own second-generation compound. Have some ready to go, and start developing them as soon as you have some idea (from the clinical trials) of what might need improvement. This could even mean introducing your improved drug earlier than usual, just to keep someone else from grabbing your market (better to sink your own drug than have someone do it for you.)
Merck is known for devoting a lot of effort to backup projects. Other companies don't seem to put nearly as much effort to it, which has never made sense to me. A good follow-up project allows you to have the best of both worlds: you get to be the first to do something, and you get to make most of the money from it. What a deal!
I'll Have What They're Having
An article in the latest Nature Reviews - Drug Discovery from the McKinsey consulting firm calls into question a dear pharma dogma: that the real money and glory comes from being "first in class" with a new type of drug. Over the years, a number of companies have made this an official goal of their research efforts, and projects of this kind are often allotted more resources. That's because they tend to need more work, true, but they're also perceived as more valuable. After all, here's this big unmet medical need, and you're going to be the first to fill it.
The authors went back over the marketed-drug landscape and tried to figure out where most of the money was coming from. When they looked at the 32 biggest blockbuster drugs from the last ten years, fewer than one-third were novel first-in-class therapies. An equal number were defined as "late-comers," coming in more than fifteen years later than the first drugs of their kind. Out of the 189 drugs launched by the top 15 companies in the last ten years, the group with the highest value (measured by sales) were in between these two: the "fast-followers," coming in between 2 and 15 years after the first in their class.
So, what does this mean? That the industry is full of me-too tagalongs? You can find plenty of people with that view, and some of them get rather red in the face when they talk about it. But the thrust of this article is that the drug industry is being too innovative for its own financial health. Now there's a contrarian point of view!
To be sure, I think that some of the outrage over "me-too" drugs comes from people who have, perhaps, an unrealistic view of how difficult it is to come up with the truly new ones. These folks start from the idea that most everything that hits the market is (or should be) new - why else would it be coming on the market? So the reality proves, well, disappointing. The thing is, any company that tried to do only first-in-class drugs would have gone out of business a while back. They're just too hard to find, and too hard to do, and I'd argue that they're getting harder in both categories.
As the current paper shows, though, later drugs can compete well if they have something to differentiate themselves by. The authors argue that the traditional criterion, efficacy, isn't always the best one. Safety, dosing schedule, or a greater breadth of disease indications are all alternatives. Trying to compete on too many differences at the same time has a higher risk of failure. The authors recommend picking one good difference and pounding away on it. Another mistake is to try to come in with a follower drug that has some area in which it's actually worse than the existing competition - no matter what it has going for it otherwise, a real deficiency is going to be hard to overcome.
I think that this article suggests some corollaries: first, that over time, it gets harder and harder to have a big drug in a given category. The bar has been raised by each earlier entrant (if the companies involved knew what they were doing, that is.) But this takes place at different speeds in each therapeutic area, so just counting years or drugs isn't enough. Look at Lipitor - even Warner-Lambert's own people didn't think that there was room for another statin, but man, was there ever. Secondly, though, the clock does have to run out at some point. I wonder if it hasn't finally done so for the statins. Will AstraZeneca's Crestor take over, or has it come too late in the game? I tend to think the latter, but keep in mind that I would have said the same think about Lipitor. . .!
There are also some things to say about the strategies a drug company should pursue. I'll talk next week about the timing of these drugs, and how some of them haven't worked that well. It's also worth talking about just how much of your company's time should be spent on the new stuff, and how much on ripping everyone else off. How about all-me-too, all the time? Would it work? More to the point, as far as I'm concerned, would anyone in research want to work there?
We'll talk about that one, too. See everyone on Monday!
A Few Quick Ones
Not much blogging time this evening - I have pressing business the next day or two: getting things ready for a five-year-old's birthday party! I'll have more comment later on the chemistry Nobels, which I clearly didn't see coming. (That's probably because I didn't consider these discoveries to be all that chemical, and I have to wonder if the awardees consider themselves chemists, either.) These were worthy discoveries; it's just that I would have picked them for an award in the Physiology/Medicine category. But what the hey, that (also well-deserved) award this year was partway towards chemistry and partway towards physics. The Nobels were rather multidisciplinary this year.
There are a few other things I've been meaning to mention. One is a new addition to the blogroll, Carl Zimmer. He's a science journalist who's just started a very interesting blog, which I can strongly recommend.
And here's an interesting rundown of Schering-Plough's prospects (pt.1, pt. 2), as worked out by a value-investing hedge fund manager. I think it's an excellent take on the subject, and I agree with his conclusions. Which is unfortunate, since I'm an owner of the stock (for now.) I'd be interested to see this guy take on one of the higher P/E drug stocks and see if its numbers make sense - I've already gone on record as saying that I can't understand Pfizer's optimism, for example.
And finally, since I was mentioning "unnatural products" as a synthetic target, I really should call attention to some of the most unnatural ones that have ever been synthesized. A professor at Rice has engaged in "anthropomorphic synthesis," making molecules that look like stick figures of people. His aim (and his funding) for this project seem to be for outreach to kids; I have to wonder if it's going to make a difference or not that there's a smear of yellow compound in the flask. Otherwise, it seems like you could have just drawn these things (although, as one of my correspondents pointed out, the way he's drawn some of them is a bit of an abuse of the ChemDraw program.) Reaction to the forthcoming paper in the Journal of Organic Chemistry, from what I can see, has not been favorable.
By the time I get to work tomorrow, we'll know a lot more about who won. No, not that business in California. The Nobel Prize in Chemistry will have been announced! Someone (or some group of up to three people) will have had early-morning phone calls from Sweden, telling them something that they'll hardly be able to believe. The only thing I'm sure of is that they won't be calling me. Well, actually, I can think of a lot of other people that they won't be calling, to a good level of certainty. But who will they call?
I have only speculation to offer, with a few hours left to be proven wrong. If there's a favorite, an obvious choice, it would be an award for genomic sequencing. It's not the first thing you think of when you think "chemistry," but the PCR award (1993) was a chemistry, so why not? This one would be Craig Venter, and you'd have to have the NIH sequencing effort recognized too, surely by naming Francis Collins. Such an award wouldn't particularly surprise or thrill anyone (except the recipients, of course!) because it's such a large landmarky achievement. Perhaps it's too large to single out individuals in this way, actually, which is an issue that seems to have been on the minds of the Nobel judges before.
I don't think the award will go to anyone in my branch of chemistry. The 2001 award went to three organic chemists, and last year's was analytical chemistry. The chemistry prize sort of pings around between those fields, and biochemistry, physical chemistry, theoretical/computational chemistry and other subdisciplines, so I think it's too soon to expect an organic award. Maybe in a couple of years. As far as I know, there's never been a chemistry Nobel that recognized medicinal chemistry - I think that committee considers it to be medicine, by tradition if nothing else.
Perhaps the committee will reach back to find something worthy that everyone thinks should have been recognized long ago. The Physics prize this year seems like a good example of that - the awardees mostly seem surprised that the committee finally got around to recognizing them; I think they'd pretty much given up hope. Sometimes I think these people end up feeling like Vladimir Nabokov, when Lolita finally made him famous and wealthy. "But this all should have happened twenty years ago," he reportedly said.
There's an article of great interest in the Sunday New York Times. Headlined "Where Are All the New Drugs?", it looks at the situation at GlaxoSmithKline, a company whose name makes clear that it was formed by a gigantic merger. The article is a clear-eyed look at the effect of such mergers on the drug industry, and whether they do anything close to what they're supposed to.
It's a topic I've visited several times over the last couple of years. As fate has (so far) had it, I haven't worked for a company that's gone through a merger. It's been a pretty close thing, though, on a couple of occasions, and those are just the ones that I know about. At one of my previous companies, continual rumors had us paired off with just about everybody short of Pizza Hut. That wasn't a big productivity booster, as you'd imagine, but it's nothing compared to what happens when a merger actually goes through.
The article makes a good point that "mergers of equals" are probably the worst. Not that the others are much fun, but when Pfizer opens wide to swallow your company, you at least have a fair idea of where you stand (for one thing, your company's new name is going to be "Pfizer.") There's a lot of disruption, but eventually the anaconda digests you and sorts you out. But when Glaxo and SKB merged, there seems to have been a prolonged period of dancing around - a sort of "After you," "No after you", but conducted with knives. Former (and current?) scientists interviewed by the Times speak of "an incredible number of arguments about who was doing what."
On top of that, all the standard procedures at both companies came under review, so not only did people not know who was doing something, they weren't sure how it should be done in the first place. A favorite quote: "I was in meetings thinking 'My life is futile.' Moving piles of dirt from here to there will make you crazy." (Actually, a long enough meeting on almost any subject is enough to make me ponder the worth of my existence, but maybe that's just me.)
And to what end? Is GSK's pipeline full, twanging with that bouncy synergy that the merger-meisters promised? Not exactly. The Times article highlights Levitra, for example, as a recent drug, but that one came from Bayer (GSK is there to provide the marketing muscle.) Otherwise, they don't have too much coming along, and a number of projects from both of the original companies appear to have gone off the rails.
To be sure, projects go off the rails all the time in this business. And I know that GSK has invested large amounts of money in some very difficult research areas (nuclear receptors, for one.) And no one's drug pipeline is all that full these days. But wasn't the merger supposed to be the cure for these ailments? The GSK defenders in the article are reduced to arguing that the merger wasn't as bad as people are painting it. Perhaps. But it was supposed to be great. What happened to that?
Performing Proteins, Jumping Through Hoops
The new way to detect proteins was published in the latest issue of Science (, p. 1884). It's an ingenious idea, because it gets around one of the biggest problems. As I was saying on Friday, proteins have a much greater variety than DNA or RNA, and can take on a gigantic range of shapes. So It's no surprise that there isn't a general technique that picks them up specifically.
The closest thing is the use of antibodies. The immune system is capable of generating a binding protein against something it recognizes as foreign, and these can be very selective. Antibodies are a vital tool in the protein field, but antibody assays don't approach the sensitivity of the DNA/RNA assays that are behind techniques like gene chips.
Over the past ten years or so, several groups have tried to combine the two approaches. Generally, what they've ended up doing is binding the target protein to some solid support, and then using an antibody for it that has a specific stretch of DNA attached to it. Once the antibody binds to the protein, you can use techniques like PCR to amplify the DNA that's hanging off of the complex, and get a much higher sensitivity in the assay. This "immuno-PCR" idea is a good one, but it's had some problems.
From a chemist's standpoint (mine!), one of the biggest difficulties is hooking up the DNA to the antibody. There's no really good way to do that, because nucleic acids and proteins aren't very similar, chemically. I mean, it's certainly possible, but it's not a natural fit. Several methods have been used, but they all have their complications. And when you can get them to work, you typically don't have many DNA strands (maybe only one) coming off your antibody, which isn't optimal.
This is a lurking weakness of many interesting molecular biology schemes, actually. I've come across several of these ideas that have a step in them which says "And then we hook our small ligand to the biomolecule," or "OK, we should be able to chemically link these two protein fragments together with some kind of nonreactive tether, right?" Joining things up like this is pretty difficult, actually. It's a special case each time, and each time you have to make an effort to be sure that you haven't hosed your system up somehow. We organic chemists don't have a toolbox marked "unreactive, uninterfering all-purpose linking groups."
But this latest wrinkle, from Chad Mirkin's group at Northwestern, gets around these problems. They start with small (micrometer-sized) polymer particles that have iron oxide in their core. These magnetic microparticles are functionalized with monoclonal antibodies (mAbs) to the protein of choice - in this case, they went after the well-known diagnostic protein PSA (prostate-specific antigen.) The monoclonals all recognize the same surface region of the PSA. Separately, they take gold nanoparticles and decorate those with plenty of specific DNA strands, as well as attaching polyclonal antibodies for the PSA protein. Both of these technologies have been previously worked out. The polyclonal antibodies recognize PSA, too, but at a number of different sites that don't necessarily overlap with the one that the mAb sticks to.
So, you take your magnetic particles, with the mAbs all over them, and let them contact your blood or tissue sample. PSA proteins stick to the antibodies, as they should. Now you purifiy things by use of the magnetic microparticles - put a strong magnet up there, and all the microparticle/antibody/PSA complexes stick to the wall of the tube! That lets you wash out all the other stuff - other proteins, unreacted PSA, what have you.
Then you come in with your gold nanoparticles, and let the polyclonal antibodies stick to other exposed parts of those bound PSA proteins. Now you've got some real hairballs floating around in there - micrometer-sized magnetic particles, decorated with antibodies which have PSA protein stuck to them, which in turn have other antibodies stuck to them, which antibodies are attached to small gold particles that are furry with strands of DNA. Quite a fearsome sight, no doubt, if we only we could see it. Another magnetic-separation round can clean up this reaction, too.
But now things are set up. Soaking the complexes in pure water allows the DNA to unravel off of them. You use the magnetic separation again to get rid off all the hairballs, which have done their part and are dismissed. That leaves you with a water solution of single-stranded DNA, which (thanks to PCR and cDNA chips) can be detected at laughably low concentrations. The linkage of protein-to-DNA has been accomplished smoothly and cleanly, making for a very nice assay indeed.
Compared to the standard diagnostic PSA method, this one is about a million times more sensitive. And there's more: by using a mix of antibody-coated particles, you could simultaneously detect dozens of proteins in the same sample. Just assign each of them a specific DNA sequence back at the beginning, and they won't interfere with each other at all. The authors call this "bio-bar-coding," and that's a good way of looking at it. This should work for any protein that antibodies can be raised to, which is a goodly selection (and which covers most everything that's diagnostically important, as far as I'm aware.)
I assume that these folks have already patented this idea - heck, I assume that they've already formed a company, selected a letterhead and, most likely, a ticker symbol. If this works out as well as it looks, it's going to be very, very useful, both in the clinic and in the research lab. Congratulations all around.
Horn, Dept. of Blowing Own
To my surprise, Forbes magazine has picked this site (as well as fellow Corantean Richard Gayle's Living Code as one of the best medical weblogs. The others on the list are well worth reading, of course, and there are plenty of medical blogs that I think could easily have been included. I'm not sure what Richard and I are doing in there with a bunch of MDs, but I'm certainly happy for the mention!
Proteome on the Range
One of my sidelines here at Pipeline HQ (besides the one of, er, working for a living in the drug industry) is writing a monthly column for a trade magazine called Contract Pharma. Reading my column is pretty similar to reading this blog, but since it's written for a narrower audience, I can leave out some background information and thus snark around more quickly in a given space.
My most recent column included a mention of the field of proteomics, one of the many similarly-named concepts that are taking over the research world. (I realized things had come to a head when I saw an ad for a new journal called - wait for it - Omics.) Proteomics is the word used for attempts to figure out the whole world of protein-protein interactions in living systems, a huge (and hugely worthy) goal. It's the successor to genomics, which was all the rage three or four years ago, but now makes people shift around uneasily and look at their shoes. (Richard has a good piece on why that is over on Living Code.)
But there have been a number of problems in the proteomic world. As I mentioned in that recent column, a big one is that there's no protein equivalent to the wonderful ability of DNA and RNA molecules to recognize each other. For those, you can make "gene chips", packing them with short single-stranded nucleic acid sequences, and they'll pick out their complementary partner sequences with amazing speed and accuracy. Then there's PCR: the polymerase chain reaction will take a sequence of DNA and copy out huge piles of it with extremely high fidelity, which is just the sort of thing you need to boost the signal / noise of your assay.
Proteins don't do any of that. It's not their job, and they haven't been tuning up for a billion years to do those useful things for us. Their functions, while spectacularly useful, don't lend themselves to the same kind of exploitation. On top of that difficulty, we have, honestly, only a rudimentary understanding of why some of them bind to each other. Computer modeling of these interactions is not for the sqeamish, those short of funds, or those pressed for time. The interaction of those complementary DNA sequences is unstoppably strong, but while proteins are capable of tenacious binding, they sometimes just seem to sort of graze each other.
For the same reasons, proteomics people can only look longingly at things like PCR. There's no fast or easy way to make more of any given protein. (If there were such a method, Roche and Trimeris would be using it to make Fuzeon, Lilly would use it to make insulin, and so on.) You can purify it from the living cells, which is often a specialized job. Or you engineer a cell to make it, which is a standard trick, but it isn't always possible. Sometimes makes the purification problem even worse. Or you really bite down and synthesize it from scratch, a la Fuzeon, which will cost you serious time and money for any decent-sized protein.
One reason things are so messy is that proteins have some twenty variable amino acid building blocks, as opposed to four nucleic acid bases. So their potential for variation can give you the shivers: let's see, I'll make a library of short peptides, say, eight amino acids long. Surely those are all known by now. That's an easy job for a peptide synthesizer machine, at that length. . .why not make 'em all? What a screening tool that'll be! Twenty milligrams each, that'll keep my assays running for years. Now, we've got, hmmm, twenty to the eighth, that's. . .arrgh, that's over twenty-five billion peptides, is what it is. And at twenty milligrams apiece, that's 512,000 kilos of material. A little under five adult blue whales worth. Urk.
Larger libraries get even more exponentially insane. Sticking with the 20-milligram vials, at between 22 and 23 amino acids, you're past the weight of the Earth, assuming that it's all been transformed somehow into little twenty-mg piles of peptides. At 27 amino acids, your library of proteins weighs more than the sun. Insulin, to give you a size comparison, is 51 amino acids long. Clearly, some serious tools have to be developed to deal with these things. And a serious one may just have been, which I'll talk about on Monday.
Time to follow up on my post on natural products. These structures, mainly because they're so weird, definitely fight above their weight in the world of organic synthesis. Some of the reasons are historical. Back in the days before modern spectroscopy, the only way to be sure of a complicated structure was to synthesize it, using reactions that you trusted. That rationale eroded during the 1960s and 1970s, mainly because of the rise of high-magnetic-field NMR machines (and tricky ways to extract ever more information from them.)
There's almost no place left for synthesis as a proof of structure. It's hard now for chemists even to realize what it was to live in a world where it was needed; it's like contemplating having to hunt and forage for all your food. Nonetheless, people still try to make natural products. For one thing, they've been traditional proving grounds for new reactions ("Well, OK, but can you actually makesomething with it?") I can see the point, but I don't think that a reaction is somehow more real after it's been used to make something that came from a liverwort. Some of this attitude comes from the challenge of trying to duplicate nature. That's what I was getting at when I spoke enviously about the bacterium that made the molecule I spent most of my graduate career attacking.
And there's another rationale for synthesis that usually comes up a lot sooner than I've been letting it: the biological activity of many of these structures. That's always been a big mover of the grant money, if your favorite wild structure is active against cancer (like taxol or esperamycin.) I, too, hewed to this line when I was in graduate school. My molecule was an antibiotic, and although it was clear that any synthesis we achieved was going to be hideously uneconomical, the idea was that the work could produce some useful new compounds.
Would that it had. The truth is, most of these syntheses are so arduous that they're never used to generate analogs. Near the end of the synthesis, there's generally so little compound available that nothing could be done with it, anyway. There are honorable exceptions, but most natural products are tweaked for research by starting from the natural product itself (isolated from the organism) and dinking its functional groups around from there. And if you can't get enough of it to do that, then it's never modified at all.
So what are we left with? People like to point out that total synthesis does a great job of training graduate students. And that's true, but there are a lot of other things that train them, too. My feeling has long been that if you can put up with total synthesis work, you can put up with anything, so it does have some value as a selection too. But I've seen just as many good chemists in industry who did other kinds of organic chemistry.
No, the real reason that many of these molecules are tackled is because they're there. And because they look impossible, and because there's prestige associated with being able to make them (and especially to make them first.) Don't forget that these syntheses take a tremendous amount of work from many hands, so some of the stature comes from being able to have a huge research group and to motivate them to work days, nights, weekends, and national holidays.
But keep in mind that there would be no prestige at all in leading a similar effort to make something equally bizarre that you just dreamed up. On the contrary, people would think that you were nuts. There are very few exceptions - Leo Paquette's synthesis of the "unnatural product" dodecahedrane comes to mind, and there were people who grumbled about that, too. Of making crazy structures there is no end: ten minutes and a little imagination, and you can draw something that it would take twenty people five years to actually make. Increasingly, I'm having trouble seeing the difference between that little thought experiment and total synthesis as it's practiced today.
I think it's a decadent area of chemistry. The monuments it produces are huge, complicated, and impressive, true - but we've already proven, over and over, that we can produce huge and impressive monuments. I don't think there's a molecule now that can't be synthesized, given sufficient money, time, and frantic post-docs working 18-hour shifts. Does anyone in chemistry really think differently? Be honest, now. So why are we still doing this stuff? What's there left to prove?
Postscript: Greg Hlatky at A Dog's Life has a good post on this subject. Also, I know that I have readers at universities that do a lot of this sort of work. I mean no offense to any of you folks at the bench, who are trying to finish a huge molecule, and took a break from it only to find me putting down the whole enterprise. Hey, I was there, too. It's your principal investigators that I might have some issues with. Their names wouldn't mean anything to people outside the field; and those in it already know who I'm talking about!
Archives, RSS & Email
Click here for access to our archives.
Dictionary of Scientific Quotations
If the Universe is Teeming with Aliens
(Then where is everybody?)
The Night is Large
A Martin Gardner collection
Eurekas and Euphorias